I guess I'm not understanding your argument; seems to me that polling actual voters is the best way to get a handle on the characteristics of actual voters. Are you proposing there's a big enough change from election to election in the composition of persons who actually vote (one way or another), so extrapolating from one election to the next isn't valid? If you have a source showing that, I'd be interested to read it.
So no, its not perfect - there's always going to be some shift, but I don't agree that the change is likely to be so huge over 4 years that the cited numbers are unpredictive of where we are in '04.
No, I'm not arguing with the poll results; I understand that the methodology is generally sound, and the trends pointed out generally ring true.
However, in going over the article again, I find that I have unanswered questions: Who did Ipsos-Reid poll to discover that "in pro-Bush counties only 7 percent of voters earned at least $100,000, while 38 percent had household incomes below $30,000"? Eligible voters? Registered voters? People who actually voted? Are these exit polls? Were they conducted in person or by phone? And given the weird reporting that went on during the election itself, if I-R did conduct exit polls, how accurately do those polls reflect the number of votes tallied?
I think I'm particularly skeptical because of the source; The American Enterprise has a very clear agenda and is not above twisting numbers to fit their agenda. Not that TAE is unique in this regard. Lots of publications on both sides of any given issue are prone to quoting statistics that seem to favor their side over the other side. When I see stats on Mother Jones suggesting that Donald Rumsfeld eats puppies, I'm equally sketical.
Do you, perchance, have a link to the I-R poll in question? My google-fu has failed me.
Apparently the sample was "3,161 registered voters" nationwide, done in 2002. As a methodologist, my concern would be with where they got their income numbers from; traditionally, income shows up as one of the questions lots of people lie to pollsters about. I'm guessing they used Census 2000 income data at the county level and crossmatched it with the I-R poll results, since they apparently weren't able to do a straight "Bush voters reported earning this much, Gore voters this much" comparison.
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
timeasmymeasure
no subject
Date: 2004-09-14 05:06 pm (UTC)I guess I'm not understanding your argument; seems to me that polling actual voters is the best way to get a handle on the characteristics of actual voters. Are you proposing there's a big enough change from election to election in the composition of persons who actually vote (one way or another), so extrapolating from one election to the next isn't valid? If you have a source showing that, I'd be interested to read it.
So no, its not perfect - there's always going to be some shift, but I don't agree that the change is likely to be so huge over 4 years that the cited numbers are unpredictive of where we are in '04.
no subject
Date: 2004-09-14 09:16 pm (UTC)No, I'm not arguing with the poll results; I understand that the methodology is generally sound, and the trends pointed out generally ring true.
However, in going over the article again, I find that I have unanswered questions: Who did Ipsos-Reid poll to discover that "in pro-Bush counties only 7 percent of voters earned at least $100,000, while 38 percent had household incomes below $30,000"? Eligible voters? Registered voters? People who actually voted? Are these exit polls? Were they conducted in person or by phone? And given the weird reporting that went on during the election itself, if I-R did conduct exit polls, how accurately do those polls reflect the number of votes tallied?
I think I'm particularly skeptical because of the source; The American Enterprise has a very clear agenda and is not above twisting numbers to fit their agenda. Not that TAE is unique in this regard. Lots of publications on both sides of any given issue are prone to quoting statistics that seem to favor their side over the other side. When I see stats on Mother Jones suggesting that Donald Rumsfeld eats puppies, I'm equally sketical.
Do you, perchance, have a link to the I-R poll in question? My google-fu has failed me.
Thanks!
no subject
Date: 2004-09-14 10:15 pm (UTC)I think the poll that article was referring to is this one:
http://nationaljournal.com/about/njweekly/stories/2002/0315nj1.htm
http://www.ipsos-na.com/news/pressrelease.cfm?id=1453
Apparently the sample was "3,161 registered voters" nationwide, done in 2002. As a methodologist, my concern would be with where they got their income numbers from; traditionally, income shows up as one of the questions lots of people lie to pollsters about. I'm guessing they used Census 2000 income data at the county level and crossmatched it with the I-R poll results, since they apparently weren't able to do a straight "Bush voters reported earning this much, Gore voters this much" comparison.
no subject
Date: 2004-09-14 11:38 pm (UTC)Thank you for the link.